method detection limit (mdl) development and … detection limit definition “the method detection...

Method Detection Limit (MDL) Development and Standardization

2009 National Ambient Air Monitoring ConferenceNashville, TN

November 4, 2009

OverviewWe’re all using 40 CFR APPENDIX B TO PART 136 —

DEFINITION AND PROCEDURE FOR THE DETERMINATION OF THE METHOD DETECTION LIMIT—

REVISION 1.11

Method Detection Limit Definition

“The method detection limit (MDL) is defined as the minimum concentration of a substance that can be measured and reported with 99% confidence that the analyte concentration is greater than zero and is determined from analysis of a sample in a given matrix containing the analyte.”*

* 40 CFR Appendix B to part 136

Method Detection Limit Working Definition

Statistically calculated concentration where you would expect to “qualitatively”

identify the target analyte

Measure of how well you can repeat an analysis

Function of the ability to prepare identical low concentration samples.

Practical MDL Determination

Your MDL is a measure of YOUR laboratory’s sensitivity using YOUR chemicals, equipment, and staff.

If you spike at your MDL concentrations you should find:

50% of the values would fall above the MDL (detected)

50% would fall below (not detected)

MDL Basics1.

Analytical systems (Instruments)

Run on systems that are operating properly

Calibration meets criteria

Columns in good condition

Avoid contaminate carry-over from previous samples

Blank samples meet criteria

Routine maintenance is complete

MDL Basics, Continued

2. Calibrating for the MDL procedure

MDL studies are typically done at the beginning of a season or new year.

Generate a new calibration curve prior to analyzing the MDL samples.

Calibrate using the same calibration range used for field samples

Verify calibration curve with second source standard

MDL Basics Procedures to Improve the MDL

3. Choosing the proper spike level !

Prepare standard 2.5 to 5 times the estimated detection limit. (MDL is a function of the spike concentration!)

Analyze at least seven (7) samples at the spike level, calculate the MDL

Accept the MDL if the calculated value is less than the spiked value.

Reprepare at a lower level and rerun the seven set series if calculated MDL is greater than 5 times the spiked level

Calculated MDL < Spike Level < 5 x Calculated MDL

Procedures to Improve the MDL, Continued

4. Replicate sample preparation

Method requires at least 7 replicates –

ERG recommends 10

Ensures the minimum number (seven) of replicates are met in the event of outliers, which should only include:

Obvious analyst error

Improper sample preparation

Reject entire outlier sample data set

Use the correct Student’s T value for the number of replicates (n-1 degrees of freedom)

Procedures to Improve the MDL, Continued

5. Analyzing blanks

Analyze at least one method blank to measure background contaminations

Minimizing the blank helps control the variation (precision) of the MDL replicate runs

With the exception of metals, blank subtraction is not allowed for MDL determination.

Calculations to Determine MDL

Three important things to remember about calculating MDLs are:

Use the sample standard deviation, Use the correct Student's t-value, and Use correct significant figures

Calculation 40 CFR Appendix B part 136

Number of replicatesDegrees of Freedom

(n-1) t(cn-1,.99)

7 6 3.1438 7 2.9989 8 2.89610 9 2.821

ExampleMDL=2.821 (Spooled

)where 2.821 is equal to t(10,1-α=.99)

MDL Verification

Analyzing a single sample spiked at the MDL concentration

If the analytical response is NOT distinguishable from a reagent blank, the calculated MDL is unreasonably low

This could happen if you make exact replicate samples and your system is inherently noise free.

The MDL study should be repeated at a different concentration. (Higher or Lower?)

If the analyte is detected at the presumed MDL, the MDL is defensible and should be reported

MDL Issues: What Affects Precision

Background interferences and Blank contamination are variable and raise MDL spike

Precision of standards preparation equipment (volumetric glassware, gas metering equipment, syringes etc) is variable

Physically unable to produce a low enough standard to perform MDL study

Instrument noise

Others?

MDL Issues: What Affects TO-15 Precision

VOC concentrator performance

Different behavior of polar vs. nonpolar TO-15 compounds

Variation in canister manufacture, use, or age

Precision of standards preparation (mass flow controllers, etc)

Ability to make standards concentration low enough to perform MDL study

Others?

Spike Sample Preparation for Canisters (TO-15)

Static Dilution

Spike the canister with a mixture of liquid components prepared in static dilution bottles

Dynamic Dilution

Mix standards and humidified zero air with mass flow controllers and a calibration manifold

Better precision and lower detection possible with dynamic dilution spike preparation.

MDL Issues: What Affects Carbonyl Precision

Background and blank contamination: Interferences from DNPH cartridges are variable and raise MDL

Precision of standards preparation equipment (volumetric glassware, syringes, etc)

Carbonyl extraction technique

Others?

Spike Sample Preparation for Carbonyls (TO-11A)

Vendor prepared stock solution

Prequalified cartridge blank Lot

High quality solvents

Class A glassware

Gas tight syringes

Repeatable spiking technique

For determination of Quartz filters that have a high background:Analyze to initially determine which elements

have background interferences

Spike filters and determine the MDL using standard procedures for elements w/o background

Analyze 7-10 non-spiked filters to determine MDL for filters that have high background

MDL Issues: What Affects Metal Precision

IO-3.5 MDL Example Analyte

Average Quartz Filter (ng/strip)

Antimony 6.4Arsenic 11.5Beryllium 4.1Cadmium 23.3Chromium 408Cobalt 6.0Lead 77.7Manganese 93.3Mercury 11.7Nickel 43.6Selenium 10.7

IO-3.5 MDL Example (cont.)

Analyte

Average Blank Quartz

Filter (ng/strip)

Spiked Amount

MDL (ng/strip)

Chromium 408 BLANK 76.2Nickel 43.6 BLANK 29.4

Analyzed a spike concentration at 25 and 75 ng to verify MDL concentration.

Summary Improving MDLs

Control variation in spike sample preparation

Control background as much as possible

Control instrument performance

Improve sensitivity of analysis

Larger injections

Concentrate samples

Sensitive detectors (e.g., Full Scan vs. SIM MS)

MDL Common Sense CheckDoes the spike level exceed 5 times the

MDL? If so, the spike level is high. Is the MDL higher than the spike level? If

so, the spike level is too low.Are the replicate recoveries reasonable?

MDL Final Check

Does the calculated MDL meet the objectives for your program?

Julie Swift Eastern Research GroupJulie.swift@erg.com919-468-7924

Contact Information